Buy data / AI companies

Second customer

Training data for AI companies

Foundation-model labs, applied AI shops, and data-label teams that need real firm names, product text, and edges — not another Common Crawl slice.

What you need

A licensed graph you can train on. Parties as filed. Product descriptions as filed. Year and port context. A document that says what you may do with it.

What we put in front of you

The trade-graph extract is the product. Name variants sit next to it. Customs-country files are for teams that want more languages and HS text, not only US ocean English.

Files for this desk

Questions

What training data can I buy for a supply-chain model?

A licensed shipper–consignee–product graph from US ocean bills, plus the name-variant clusters from the name-fragmentation paper. That is not Common Crawl. It is filed names, product text, and edges.

May we train a model and keep the weights?

Yes, under the license on the quote. Redistributing the raw graph is not included unless we write that in.

Other desks: Universities Finance and trade desks All files

Request this file

Standard processing time is 3 to 7 days after we confirm the buy. Say the years, the country, and whether you need a sample header first.